There are various known approaches to upgrading nodes in a cluster, for example updating software on such nodes.
One known approach is to stop the entire cluster, upgrade its nodes, and then restart the cluster. This approach suffers from there being a period of time where the cluster is unable to provide any services.
Another known approach is to carry out the upgrade, for example update software, on all of the cluster nodes and, once complete, have one of the cluster nodes coordinate the upgrade of all nodes to run the updated software. However, this approach requires the updated software to be installed on all of the cluster nodes before any one of them can be upgraded. Furthermore, the coordinated upgrade is, itself, a complicated process.
It would therefore be desirable to provide improved measures for use in the upgrading of nodes.